[PATCH 5 of 6 V3] match: avoid translating glob to matcher multiple times for large sets

[Boris Feld]

# HG changeset patch
# User Boris Feld <boris.feld at octobus.net>
# Date 1542916922 -3600
#      Thu Nov 22 21:02:02 2018 +0100
# Node ID b41a4db20b0898e57f9fa0bb9ebdabb30a53df75
# Parent  4fa131d8a4d7d2aea452994458eb088e9e997df4
# EXP-Topic perf-ignore
# Available At https://bitbucket.org/octobus/mercurial-devel/
#              hg pull https://bitbucket.org/octobus/mercurial-devel/ -r b41a4db20b08
match: avoid translating glob to matcher multiple times for large sets

For hgignore with many globs, the resulting regexp might not fit under the 20K
length limit. So the patterns need to be broken up in smaller pieces.

Before this change, the logic was re-starting the full process from scratch
for each smaller pieces, including the translation of globs into regexp.
Effectively doing the work over and over.

If the 20K limit is reached, we are likely in a case where there is many such
glob, so exporting them is especially expensive and we should be careful not
to do that work more than once.

To work around this, we now translate glob to regexp once and for all. Then,
we assemble the resulting individual regexp into valid blocks.

This raises a very significant performance win for large `.hgignore file`:

Before: ! wall 0.153153 comb 0.150000 user 0.150000 sys 0.000000 (median of 66)
After:  ! wall 0.059793 comb 0.060000 user 0.060000 sys 0.000000 (median of 100)

diff --git a/mercurial/match.py b/mercurial/match.py
--- a/mercurial/match.py
+++ b/mercurial/match.py
@@ -1185,6 +1185,7 @@ def _buildmatch(kindpats, globsuffix, li
         return regex, lambda f: any(mf(f) for mf in matchfuncs)
 
 MAX_RE_SIZE = 20000
+_BASE_SIZE = len('(?:)') - 1
 
 def _joinregexes(regexps):
     """gather multiple regular expressions into a single one"""
@@ -1203,21 +1204,31 @@ def _buildregexmatch(kindpats, globsuffi
     OverflowError
     """
     try:
-        regex = _joinregexes([_regex(k, p, globsuffix)
-                               for (k, p, s) in kindpats])
-        if len(regex) > MAX_RE_SIZE:
-            raise OverflowError
-        return regex, _rematcher(regex)
-    except OverflowError:
-        # We're using a Python with a tiny regex engine and we
-        # made it explode, so we'll divide the pattern list in two
-        # until it works
-        l = len(kindpats)
-        if l < 2:
-            raise
-        regexa, a = _buildregexmatch(kindpats[:l//2], globsuffix)
-        regexb, b = _buildregexmatch(kindpats[l//2:], globsuffix)
-        return regex, lambda s: a(s) or b(s)
+        allgroups = []
+        regexps = [_regex(k, p, globsuffix) for (k, p, s) in kindpats]
+        fullregexp = _joinregexes(regexps)
+
+        startidx = 0
+        groupsize = _BASE_SIZE
+        for idx, r in enumerate(regexps):
+            piecesize = len(r)
+            if (piecesize + 4) > MAX_RE_SIZE:
+                raise OverflowError
+            elif (groupsize + 1 + piecesize) > MAX_RE_SIZE:
+                group = regexps[startidx:idx]
+                allgroups.append(_joinregexes(group))
+                startidx = idx
+                groupsize = _BASE_SIZE
+            groupsize += piecesize + 1
+
+        if startidx == 0:
+            func = _rematcher(fullregexp)
+        else:
+            group = regexps[startidx:]
+            allgroups.append(_joinregexes(group))
+            allmatchers = [_rematcher(g) for g in allgroups]
+            func = lambda s: any(m(s) for m in allmatchers)
+        return fullregexp, func
     except re.error:
         for k, p, s in kindpats:
             try: